Manuscript by Fangqun et al studies nucleation in the stratospheric condition by evaluating different nucleation schemes against CLOUD laboratory measurements and ATom observations in the southern hemisphere stratosphere. Study showed that in the recently developed kinetic nucleation model nucleation rate is much better in agreement with measurements compared to widely used binary homogeneous nucleation. This paper is exactly the type of research that modelers of stratospheric aerosols need, especially if you are studying stratospheric aerosol injections. Good indicator for excellent study is that after reading the paper, you want to take something from it to your own research. In this case it is the nucleation scheme. 

Manuscript is well written, and when questions came to my mind, those were answered in the following lines. Or at least as well as possible. Overall this is an interesting and excellent study and it is difficult to find any major or even minor issues. I have just few very minor comments which would clarify some specific points:

Model and data section: 

It would be interesting to know how competition between nucleation and condensation for sulfuric acid vapor is done in  GC-APM. 

P6 L206, After reading the abstract and introduction I got the impression that in this paper only the BHN of Vehkamäki and new BIHM scheme are evaluated. This line is the first time where BHN of Yu et al. (2020) is mentioned. I think it would be good to somehow introduce this nucleation scheme earlier (just by 1 or few line(s)) and say with just few words how it differs from the BHN of Vehkamäki (new look up tables(?) I assume).

P6 Figure 1 (and later figures): Really minor thing but it would be more clear if (a) and (b) were upper left of the panel and not in the title and that order would be a-b-c in upper panels and not a-c-e as in figs 2-3. 

P7 L246 and later in the text, at least I am not familiar with “std” in units. Could this be opened up? 

P7 L251 and L253, brackets in [H2SO4]

P11 L349 You could clarify that here you refer to measurements even though it is obvious after reading the next couple of lines.

P12 L361-368 I have been struggling with this and I am not sure if the standard deviation in the largest model is surprisingly large or is it even large or not. Number concentration in AccuM2 is much lower compared to AccuM1 and Aitken modes and thus in the linear scale the size range of the standard deviation bar is not actually large as it seems to be in logarithmic scale. 

Figures 5 and 6 and captions: There abbreviation of  particle number size distribution is “PSD” while in the text it is “PNSD”

P13 L389-391 I think this is kind of expected as there is lower nucleation rate in BIMN compared to BHN_V2002, there is more available sulfuric acid vapor for condensation. However, the BHN-V2020 line is lower than BIMN and BHN_V2002 regardless of the size of the aerosol, which is interesting. It means that particulate sulfate (+small amounts of some other species) burden is lower in BHN-V2020 compared to others. Where is this “missing” sulfate in BHN-Y2020? In the gas phase, or in some other location, or removed from the atmosphere? If you could easily give an answer to this, it would be interesting to know. 

P13 L408 I find this bi-mode structure of accumulation size region really interesting. What do you think, should this be taken into account in modal aerosol schemes especially in stratospheric conditions and add one extra mode between usually used accumulation and coarse mode? Your results are not the only case which would speak for it. Of course then in the modal scheme bi-mode structure would be artificial, as it still remains unknown what is causing it.

P14 L439 Is there bi-mode structure of accumulation mode in a volcano perturbed atmosphere? I would say that there is the coarse mode and one accumulation mode. 

This is really interesting.  I like the combination of modeling, field experiment measurements, and chamber measurements.  Getting a chamber to cooperate with stratospheric conditions is no small feat.  There are some important gaps in the study that I’d like to see resolved, mainly having to do with the applicability of your datasets.

More specifically, you show that these new nucleation schemes better match observations.  But your observations do not match what is usually thought of as hypothetical SAI conditions.  This introduces a potential source of error in your study that is not well discussed.

Comments:

I’d like to see you discuss volcanic eruptions more.  Presumably if you’re coming up with new assessments of past modeling of SAI it would also affect past modeling of volcanoes.  Did we miss something very important in our previous assessments of volcanic aerosol microphysics?  Did that affect our estimates of radiative forcing or chemistry?

The period chosen (which overlaps with ATom) doesn’t have a high stratospheric loading, and the particle size is substantially smaller than would be experienced under SAI.  Is there any reason to think that microphysical behavior will be different under SAI conditions (or volcanic conditions)?  This is exemplified in Figure 4 – while it’s clear that the updated schemes better match observed CN3 than the 2002 scheme, this is only for a narrow range of CN3 and is poorly constrained for higher CN3 numbers.

12 km isn’t very high in altitude – that won’t reach the stratosphere in many places, so the fact that your scheme better matches observations doesn’t necessarily show that it better matches observations in the stratosphere.  I would like to see more discussion on how this limitation affects your conclusions about stratospheric NPF.  You discuss some of this in Section 3.2, but I’m having trouble interpreting the applicability and limitations of your study.  Relatedly, on lines 233-234, which volcanic event and how much SO2?

I’d like to see more description about the chamber.  There is more to the stratosphere than just cold temperature – one needs to include low pressure, harsh radiation, composition, etc.  Are you actually reproducing stratospheric conditions or just stratospheric temperatures?  And if the latter, how relevant are your conclusions for stratospheric NPF?

Figures 2, 3, and 5:  I don’t have a good sense for which scheme gives you better answers.  What are these “supposed to” look like?

You make a good case for a second accumulation mode.  But there are many schemes (both modal and sectional) that take a second accumulation mode into account.  Perhaps they don’t get the processes correct that would create such a mode, but they do have it.  It might be useful to point out what those schemes are doing wrong.

You could do a bit more work (or some discussion) to characterize your uncertainty.  On lines 392-431 you discuss several sources of potential error, including missing processes or uncertainty in nucleation rates.  Do you have a sense as to whether these sources are dominant or secondary?  If the former, your results are at the risk of being made obsolete by someone who addresses those other sources of error.

Very good work! The paper compares the nucleation process, between model results with two different nucleation schemes and CLOUD/ATom measurements, under the stratospheric condition. It’s nice to see the authors relate the study to the SAI simulation, which could help to improve the SAI modeling accuracy. The overall reasoning in the paper is solid and well-justified.

I have some minor comments and corrections:

Line 81-83: Providing only two publication examples (i.e., Weisenstein et al., 2022, Laakso et al., 2022) seems not enough to prove that “BHN_V2002 has been used in most SAI modeling studies”. It would be more convincing if the authors can tell us how many models use BHN_V2002. For example, there are many models involved in GeoMIP (Kravitz et al., 2013), it would be helpful if the authors can tell the GeoMIP community how many GeoMIP simulations use BHN_V2002 for nucleation simulation.

Line 136: 4°x5° is too coarse. If possible, please repeat the simulations in 2°x2.5°. If not, the authors should discuss how much the grid resolution may influence the difference between the model results and observations, especially for the comparison between model results and ATom observations at one site in Figure 6.

Line 274: why the tropics are selected as “(0°S-30°S)” instead of “(30°S-30°N)”?

Line 296: Figure 3 needs to be optimized:

(1) Set shared x or y axis label among figures (a) to (f).

(2) Adjust the location/size of figures (g) and (h).

(3) There is a horizontal dashed line on the top of the figure (h), which should be deleted.

Line 296: for Figure 3 (g),

(1) is it a coincidence that three solid lines end up with a similar nucleation rate (about 0.02 std. cm^-3 s^-1) at approximately 17.5 km?

(2) why there is an elbow point (at around 20 km) in the red solid line? In another word, why does the nucleation rate from BHN_V2002 has a much larger changing rate with height above 20 km, compared to below 20 km?

Line 319: “BHNV_2002” should be “BHN_V2002”.

Line 390-391: I think that the competition between nucleation and condensation mentioned by Laakso et al. (2022) might be a complement to the “nonlinear process” (Line 390-391) mentioned by the authors.

Line 409: I don’t understand the sentence: “Finally, the observed PNSDs show a clear AccuM2 in all seasons except Fall but the model does not predict the existence of the mode at all.”

Based on Figure 6, the model may underpredict the AccuM2, especially in summer. But we cannot say “the model does not predict the existence of the AccuM2 mode at all”.

What’s more, the authors say “the model-simulated AccuM2 standard deviations are larger in SH Winter and Spring but are smaller in SH Summer and Fall” in Line 505. If “the model does not predict the existence of the AccuM2 mode at all”, there would be no “model-simulated AccuM2 standard deviations”.

Line 437: The citation (Clement and Harrison, 1992) is missed in the References. Please check and make sure all the citations in the main text are correspondingly listed in the References.

Line 475: Suggest changing “SAI efficiency” to “SAI radiative efficacy”. Radiative efficacy refers to the radiative forcing normalized by the aerosol injection rate, which is widely used in SAI studies (e.g., Dai et al., 2018).

In the discussion part, I think the authors can highlight the importance of model development for reducing model uncertainties of SAI simulations. Some other SAI-related model development work (e.g., Golja et al., 2021, Sun et al., 2022) is worth mentioning.

For the next step, I hope the authors could consider comparing the modeled aerosol radiative forcing based on the two different nucleation schemes, which could help the Solar Geoengineering community to have a clear feeling about how much can different nucleation schemes influence the SAI radiative efficacy.

The papers mentioned above:

Dai et al., 2018: https://doi.org/10.1002/2017GL076472

Golja et al., 2021: https://doi.org/10.1029/2020JD033438

Kravitz et al., 2013: https://doi.org/10.1002/2013JD020569

Laakso et al., 2022: https://doi.org/10.5194/acp-22-93-2022

Sun et al., 2022: https://doi.org/10.1029/2021MS002816